U.S. patent application number 13/933270 was filed with the patent office on 2015-01-08 for system for user psychosocial profiling.
The applicant listed for this patent is SenseGon Technologies Ltd.. Invention is credited to Omer EFRAT, Tal Yaari.
Application Number | 20150012471 13/933270 |
Document ID | / |
Family ID | 52133505 |
Filed Date | 2015-01-08 |
United States Patent
Application |
20150012471 |
Kind Code |
A1 |
EFRAT; Omer ; et
al. |
January 8, 2015 |
SYSTEM FOR USER PSYCHOSOCIAL PROFILING
Abstract
A profiling unit is provided herein. The profiling unit
comprises a statistical module that characterizes user activity
data statistically; a normalization module that normalizes the
statistical data related to each user with respect to user
populations; and an analysis unit that analyzes a correspondence
between normalized user study data and user archetypes, and also
associates, for each user, the normalized statistical data with one
of the user archetypes according to the analyzed correspondence.
The correspondence analysis is carried out by applying a heuristic
genetic algorithm on an artificial neural network that represents
the relation between the normalized user study data and the user
archetypes.
Inventors: |
EFRAT; Omer; (Tel Aviv,
IL) ; Yaari; Tal; (Bnei-Brak, IL) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
SenseGon Technologies Ltd. |
Ra'anana |
|
IL |
|
|
Family ID: |
52133505 |
Appl. No.: |
13/933270 |
Filed: |
July 2, 2013 |
Current U.S.
Class: |
706/15 |
Current CPC
Class: |
G06N 3/086 20130101;
G06N 7/005 20130101; G06Q 10/00 20130101 |
Class at
Publication: |
706/15 |
International
Class: |
G06N 3/02 20060101
G06N003/02 |
Claims
1. A profiling unit comprising: a statistical module arranged to
receive user activity data and derive therefrom a plurality of
statistical data that characterize the user activity data with
respect to a plurality of users; a normalization module arranged to
normalize the statistical data related to each user with respect to
at least one user population; and an analysis unit arranged to
analyze a correspondence between a plurality of normalized user
study data and a plurality of user archetypes, and to associate,
for each user, the normalized statistical data with one of the user
archetypes according to the analyzed correspondence, wherein the
correspondence analysis is carried out by applying a heuristic
genetic algorithm on an artificial neural network that represents
the relation between the normalized user study data and the user
archetypes, and wherein the profiling unit is at least partially
implemented in computer hardware.
2. The profiling unit of claim 1, wherein the analysis unit
comprises: a modeller arranged to represent the normalized
statistical data as the artificial neural network; a profiling
module arranged to apply the heuristic genetic algorithm on the
artificial neural network represented by the modeller; and a
trainer arranged to train the profiling module with obtained
normalized user study data.
3. The profiling unit of claim 1, wherein the user activity data
comprises at least one of: user data, user messages and user
activity, related to user activity in at least one of: at least one
social network and at least one internet forum or group.
4. The profiling unit of claim 1, wherein the at least one user
population comprises at least one of: all users, users within a
group that is related to each user, correspondents of each user and
users similar to each user under specified rules.
5. A profiling system comprising: the profiling unit of claim 1; an
application interface to at least one social network platform
arranged to obtain the user activity data therefrom and provide the
obtained user activity data to the statistical module; and a
profiling interface arranged to present the association of users
and user archetypes carried out by the analysis unit.
6. A profiling method comprising: deriving, from obtained user
activity data, a plurality of statistical data that characterizes
the user activity data with respect to a plurality of users;
normalizing the statistical data related to each user with respect
to at least one user population; analyzing a correspondence between
a plurality of normalized user study data and a plurality of user
archetypes by applying a heuristic genetic algorithm on an
artificial neural network that represents the relation between the
normalized user study data and the user archetypes; and
associating, for each user, the normalized statistical data with
one of the user archetypes according to the analyzed
correspondence, wherein at least one of: the deriving, the
normalizing, the analyzing, the applying and the associating is
carried out by at least one computer processor.
7. The method of claim 6, further comprising training the heuristic
genetic algorithm with obtained normalized user study data.
8. The method of claim 6, further comprising presenting the
association of users and user archetypes to an application.
9. A computer program product comprising a computer readable
storage medium having computer readable program embodied therewith,
the computer readable program comprising: computer readable program
configured to derive, from obtained user activity data, a plurality
of statistical data that characterizes the user activity data with
respect to a plurality of users; computer readable program
configured to normalize the statistical data related to each user
with respect to at least one user population; computer readable
program configured to analyze a correspondence between a plurality
of normalized user study data and a plurality of user archetypes by
applying a heuristic genetic algorithm on an artificial neural
network that represents the relation between the normalized user
study data and the user archetypes; and computer readable program
configured to associate, for each user, the normalized statistical
data with one of the user archetypes according to the analyzed
correspondence.
Description
BACKGROUND OF THE INVENTION
[0001] 1. Technical Field
[0002] The present invention relates to the field of user
profiling, and more particularly, to user profiling by integrated
statistics and experimental studies.
[0003] 2. Discussion of Related Art Internet resources such as
social networks and forums attract evermore users to intensively
interact with each other. Potentially, these interactions may be
used to characterize and profile the users, but appropriate and
effective methods are largely missing.
SUMMARY OF THE INVENTION
[0004] One aspect of the present invention provides a profiling
unit comprising: a statistical module arranged to receive user
activity data and derive therefrom a plurality of statistical data
that characterize the user activity data with respect to a
plurality of users; a normalization module arranged to normalize
the statistical data related to each user with respect to at least
one user population; and an analysis unit arranged to analyze a
correspondence between a plurality of normalized user study data
and a plurality of user archetypes, and to associate, for each
user, the normalized statistical data with one of the user
archetypes according to the analyzed correspondence. The
correspondence analysis is carried out by applying a heuristic
genetic algorithm on an artificial neural network that represents
the relation between the normalized user study data and the user
archetypes.
[0005] These, additional, and/or other aspects and/or advantages of
the present invention are: set forth in the detailed description
which follows; possibly inferable from the detailed description;
and/or learnable by practice of the present invention.
BRIEF DESCRIPTION OF THE DRAWINGS
[0006] For a better understanding of embodiments of the invention
and to show how the same may be carried into effect, reference will
now be made, purely by way of example, to the accompanying drawings
in which like numerals designate corresponding elements or sections
throughout.
[0007] In the accompanying drawings:
[0008] FIGS. 1 and 3 are high level schematic block diagrams
illustrating a profiling system according to some embodiments of
the invention,
[0009] FIG. 2 is a high level schematic illustration of the
information flow through a profiling system according to some
embodiments of the invention,
[0010] FIG. 4 is a high level schematic illustration of a profiling
system according to some embodiments of the invention,
[0011] FIG. 5 is a high level schematic flowchart illustrating a
formalization of information flow through a profiling system
according to some embodiments of the invention,
[0012] FIG. 6 is a high level schematic illustration of a profiling
system according to some embodiments of the invention, and
[0013] FIG. 7 is a high level schematic flowchart of a profiling
method according to some embodiments of the invention.
DETAILED DESCRIPTION OF THE INVENTION
[0014] With specific reference now to the drawings in detail, it is
stressed that the particulars shown are by way of example and for
purposes of illustrative discussion of the preferred embodiments of
the present invention only, and are presented in the cause of
providing what is believed to be the most useful and readily
understood description of the principles and conceptual aspects of
the invention. In this regard, no attempt is made to show
structural details of the invention in more detail than is
necessary for a fundamental understanding of the invention, the
description taken with the drawings making apparent to those
skilled in the art how the several forms of the invention may be
embodied in practice.
[0015] Before at least one embodiment of the invention is explained
in detail, it is to be understood that the invention is not limited
in its application to the details of construction and the
arrangement of the components set forth in the following
description or illustrated in the drawings. The invention is
applicable to other embodiments or of being practiced or carried
out in various ways. Also, it is to be understood that the
phraseology and terminology employed herein is for the purpose of
description and should not be regarded as limiting.
[0016] FIGS. 1-3 are high level schematic illustrations of a
profiling system 100 according to some embodiments of the
invention. FIGS. 1 and 3 are high level block diagrams while FIG. 2
illustrates the information flow through the system. Profiling
system 100 may be at least partially implemented in computer
hardware.
[0017] Profiling system 100 comprises a profiling unit 105 arranged
to receive extensive user data 115, messages by users and user
activities from various internet sources such as social networks
90, groups and forums and other sources, via an application
programming interface (API) 110 which is dedicated to retrieve
extensive data 115 from the relevant platforms. API 110 is also
termed sniffer or super sniffer to denote its retrieval
capabilities.
[0018] Profiling unit 105 comprises a statistical module 120, a
normalization module 130 and an analysis unit 140, either of which
may be at least partially implemented in computer hardware.
[0019] Statistical module 120 is arranged to receive user activity
data 115 via API 110 and derive therefrom a plurality of
statistical data 125 that characterize user activity data 115 with
respect to a plurality of users, e.g., of social network 90.
Statistical data 125 may be extensive and relate to various
metrics, forms and ways of quantifying user activity data 115 such
as counting messages, counting message lengths, assessing the
vocabulary used, message complexity, number of corresponding users,
duration of engagement in conversation, use of certain words or
word categories, inter-relations between messages etc. See further
details below.
[0020] In the model formalization, user activity data 115 is
represented as S.sub.i.sup.gl (FIG. 2). S is a basic discrete
accumulated identifier. Examples for S comprise a wide range of
parameters, for example, a number of messages for week, a number of
friend request, a number of stated conversations, a number of
published posts, and so forth. The index gl denotes the organic
object current aggregation level. Examples of aggregation levels
may comprise self-user, contact user, close environment,
demographic environment, etc. The index i denotes the organic data
entity instance. Examples or organic data entity instances may
comprise specific monitored user, a specific contact user, etc.
[0021] In the model formalization, statistical data 125, also
termed gross data, is represented as
G(S.sup.m).sub.i=T.sub.t<G(S.sup.k . . . l).sub.m . . . n>.
G(S) is a basic gross identifier. Examples of basic gross
identifiers comprise e.g. daily average message per contact user,
yearly engaged users, sum of received massages, average rate of
conversation initializations etc. T represents a linear
transformation type such as sum, average, count, variance, etc.
[0022] Normalization module 130 is arranged to receive and
normalize statistical data 125 related to each user with respect to
at least one user population to yield normalized statistical data
135 with respect to the population(s). The referred population may
comprise all users of social network 90 or comprise sub-groups of
users such as correspondents of each user, friends of the user,
users having similar characteristics or similar to the user under
specified rules etc. Normalization module 130 generates normalized
statistical data 135 that characterizes each user and is
simultaneously comparable between users due to its
normalization.
[0023] In the model formalization, normalized statistical data 135
is represented as
Nr ( S gl ) i = fNr < G ( S gl ) i , T var < G ( S gl + m )
1. . k > , T avg < G ( S gl + m ) 1. . k >> m , k
.gtoreq. 0 ##EQU00001##
[0024] Nr(S) is a relative normalized gross identifier. Nr values
are between 0 to 1 as they are normalized with respect to a whole
population. The relative normalized gross identifiers represent the
user's grade in each of his gross identifiers G(S), relative to his
dynamic environment. m is an aggregation level indicator, wherein
the selected level must be equal or greater than the current
level).
[0025] Analysis unit 140 is arranged to analyze a correspondence
between a plurality of normalized user study data 168 and a
plurality of user archetypes 80. User archetypes 80 may be
pre-defined according to socio-psychological criteria, and user
study data 165 may be based on socio-psychological studies 165 such
as profiling studies and verification studies which are externally
managed to yield effective profiling and archetype analysis.
[0026] Analysis unit 140 is further arranged to associate, for each
user, normalized statistical data 135 with one of user archetypes
80 according to the analyzed correspondence between normalized user
study data 168 and user archetypes 80. The correspondence analysis
yields profiling and segmentation data 175 of the users, which may
be used for different aims such as advertising and e-commerce, that
may be operated by various service providers and suppliers 95,
optionally but not necessarily in relation of social network 90
from which user data have been collected.
[0027] In the model formalization, profiling and segmentation data
175 of the users is represented as Behavioral Pattern (BP)
identifiers BP(S.sub.x) 180. Examples for BP identifiers comprise,
for example, practical, achiever, emotional, status seeker,
popular, risk avoider, explorer, persistent, etc. x denotes a
specific user instance.
[0028] Calculation of BP's 180 from normalized statistical data 135
represented as Nr's is carried out according to the following
expression:
BP x = [ i = 1 n .omega. i Nr i - j = n + 1 m .omega. j Nr j + j =
n + 1 m .omega. j k = 1 m + n .omega. k + d = 1 h .omega. d Nr d *
.omega. d y - c = 1 l .omega. c Nr c * .omega. c y ] 1 0
##EQU00002##
[0029] The symbols used denote: n--Number of positive effect vector
parameters; h--Number of positive effect vector only parameters;
m--Number of negative effect vector parameters; l--Number of
negative only effect vector parameters; y--Fixed number which
indicates the effect of a specific Nr on the formula and
.omega..sub.i denote the weights.
[0030] The correspondence analysis may be carried out by applying a
heuristic genetic algorithm (GA) on an artificial neural network
(ANN) that represents the relation between normalized user study
data 168 and user archetypes 80. Analysis unit 140 may comprise a
modeller 170 arranged to represent normalized statistical data 135
as the artificial neural network, a profiling module 150 arranged
to apply the heuristic genetic algorithm on the artificial neural
network represented by modeller 170, and a trainer 160 arranged to
train profiling module 150 with obtained normalized user study
data. The analysis may be carried out by a profiling module 150
operating on an ANN model generate by a modeller 170. Normalized
user study data 165 may be used to train the heuristic genetic
algorithm via trainer 160.
[0031] The GA is a key feature which operates at the core of the
behavioral identification technology to derive human personality
analysis from online psycho-social behavior. The GA performs a
behavioral psychological analysis of online communications and
other interactions over time in order to identify and classify
human behavior patterns. The GA is architecturally designed to
operate independently from applicative layers and uses a broad data
layer that is retrieved from social networks and is referred to as
user activity data 115 representing social interactions data.
[0032] In embodiments, the social interactions data (user activity
data 115) may be divided into four sub-layers. The first and basic
one is the basic demographic layer (as explained), the next three
micro-layers are composed of sophisticated formative gross data
retrieved from the user's social interactions across the social
network. Gross data (user activity data 115) is manipulated into
various types of measures such as sums, averages, variances etc. to
yield detailed statistical data that characterizes user activity
125. Detailed statistical data 125 is then normalized in comparison
with the user's different environments and relationship circles:
peer group, same gender same age group, close friends, all contacts
etc. to yield normalized statistical data 135. This normalization
process enables the GA to detect the relative location of the user
on the scale of a specific behavior pattern and so forth to
formulate his personality profile and clustering.
[0033] In embodiments, examples for social interactions data (user
activity data 115) may comprise user's basic demographic
information (such as the user's birth date, gender, place of
residence, homeland, education, work etc.), interpersonal
interaction data, public interaction and user to group
interactions. The latter three examples are illustrated below in a
non-limiting manner.
[0034] Interpersonal interaction is an interaction level which
includes all relevant data that can be gathered from users' chats
and/or offline messages. The GA collects the patterns and formative
characteristics of correspondences which indicate the overall
behavior pattern of the user's relationships and hence the user's
personality across time and contacts. Gross interpersonal data may
comprise data relating to relationships, conversations, messages,
missed calls, sequences of messages, words, punctuation marks,
chars repetition, common hours of interaction, initiation and
duration of interaction, in addition to specific vocabulary. The
activeness and conversation pattern of the user may be measured by
comparison to his interlocutors.
[0035] Public interaction is an interaction level which comprises
the user's public posts and/or public responses etc. Public
interaction gross data may comprise statuses, photos, albums,
shared links, comments, likes, statements of interests and hobbies,
like indications, applications activity etc. Common measures of
these data may comprise amount, frequency, length and category
distribution. The user may be measured compared to his responders
and/or other users discussing within the same post.
[0036] User to group interactions relate to analyzed user behavior
in public circles which include his contacts users and outsider
users. The GA gathers information regarding the user's "spreading
the word" abilities distribution relativity, leading abilities
relativity, frequency of user's contacts circle growth.
[0037] FIG. 4 is a high level schematic illustration of profiling
system 100 according to some embodiments of the invention. FIG. 5
is a high level schematic flowchart illustrating a formalization of
information flow through profiling system 100 according to some
embodiments of the invention. FIG. 6 is a high level schematic
illustration of profiling system 100 according to some embodiments
of the invention.
[0038] Profiling system 100 may comprise a distributed applications
scheduler 102, embodied, e.g., as a cloud based technology
scheduling manager, which is responsible of timing the GA's
different sub applications. The distributed applications scheduler
main feature is running applications in parallel. It is being done
by elastically allocating its resources among the applications
according to their requirements during run time. Thus, combining
smart load balancing engine and optimal resources distribution
ensure smoothly applications operations.
[0039] FIG. 4 illustrates API 110 as a super sniffer engine that
connects an external data source and profiling system 100 and
gathers data 115 by demand. In certain embodiments, data 115 may be
gathered by web pages crawling, XML/JSON parsing, connection with a
given external social network API etc. Data 115 may then be
organized and stored in a database to be provided to the structure
entity building process operated by statistical module 120.
[0040] FIG. 4 illustrates statistical module 120 as comprising a
structure entity builder 122 and a multi-layer gross calculator
124, also termed gross dynamic level transformation (GDLT) module.
FIG. 5 further illustrates information modeling in statistical
module 120, according to some embodiments of the invention.
[0041] In certain embodiments, structure entity builder 122
collects all the new raw data (115) stored in the database. By
reviewing the data and performing comparisons with existing
objects' data in the objects' hierarchy, structure entity builder
122 creates/updates a top-down full entities model and
relationships. These new objects are then used in further analysis
processes. Using the model representation, structure entity builder
122 constructs the basic discrete accumulated identifiers
S.sub.i.sup.gl from the raw data received from API 110 and then
provides the data to multi-layer gross calculator 124 which
constructs both basic gross data identifiers G(S.sup.gl).sub.l . .
. k and the transformation types T.
[0042] Multi-layer gross calculator 124 (Gross Dynamic Level
Transformation mechanism, GDLT) determined a correlation between
each and every variable between data sub-levels (e.g., the four
sub-layers illustrated above) and defines a list of transformation
types. Multi-layer gross calculator 124 thus overcomes inter-layer
data discrepancies. (The first stage in the GA's behavioral
patterns calculation is reading the new data into the objects
hierarchy and factorizing it into raw data which is used for simple
mathematical aggregations. The aggregated gross identifiers are not
completely correlated between the gross aggregation levels, a
problem solved by the GDLT mechanism, see below) For instance,
multi-layer gross calculator 124 may determine, with respect to
user activity data 115, average aggregation (and variance
included), normalized grading, down weighted average, up weighted
average and other non-linear transformations, etc. These
operations, including sub-layer integration yield detailed
statistical data 125.
[0043] FIG. 4 illustrates normalization module 130 in its function
of normalizing the aggregated data which was calculated by
statistical module 120, according to some embodiments of the
invention. The gross variables are being processed in a statistical
analysis process termed gross personalization and normalization
process. FIG. 5 further illustrates information modeling in
normalization module 130, according to some embodiments of the
invention.
[0044] This process may be carried out by three major modules:
statistical normalization and cleaning module 131, gross variable
personalized grade calculation module 132 and environmental normal
grade calculation module 133.
[0045] Statistical normalization and cleaning module 131 is
arranged to normalize, grade and clean gross data variables from
specific noise interferences. The normalization may be conducted
over the aggregated variables, by their source, lower levels' gross
variables list.
[0046] Gross variable personalized grade calculation module 132
adjusts the gross variables to the objectives. These adjustments
influence each of the gross variable values, according to the
relation policy of the personalization definition table. The
adjustments enable automatic learning of the user behavior pattern
of the user himself in a dynamic environment. By that, all the
measurements are internally identifying between different types of
users and co-relationships.
[0047] Environmental normal grade calculation module 133 performs
an adjustment to the wider society. This module perform grading
adjustments by the related other changing layers and objects in the
same environment (e.g., in normalized environments 135), by
pre-defined comparison parameters such as socio-demography. By this
ability, the GA can adjust its measurements to different cultures
and widthwise dynamic behavior trends.
[0048] In certain embodiments, analysis unit 140 (e.g., GA profiler
150) generates profiles and segmentation 175 using the following
psychological paradigm. A crucial phase in the relationship
profiling process is the analyzed user object profile itself.
Examples for profile parameters that are being collected for this
object, meaning that the GA learns the user's personal behavior
patterns: aggressiveness/passiveness, initiating habits, the way he
or she co-responds to external engagements. The relationship
parameters are generated for different data entity levels, time
scopes and environments. The relationship parameters of a single
product entity for instance, defines a time stamped position of the
two parties in every relationship signature. Besides the immediate
composition of these parameters, the GA internally stores the
incline, trend line, and other derived instances.
[0049] The relationship parameters described above are composed
from certain gross variables, arranged in a specific method. This
closed group of gross variables is pre-defined in a GA-Trainer 160
by the supervised human-controlled process (in the ANN Matrix1
process). The building method of each relationship parameter is
calculated in GA-Trainer 160 by the supervised process as stored in
ANN Matrix1 156 (e.g., in an ANN database 155, and see FIG. 6).
[0050] GA Trainer 160 is responsible of improving the GA's
calculation formulas. Given the personalization normalization and
normalization environments data (ANN Data 155), GA Trainer 160
transforms the BP engine to an automated self-learning mechanism.
GA Trainer 160 takes BP expected results combined with NR data in
order to mutate new BP formulas. The new BP mutated formulas are
created in order to improve GA's BP calculation results and reduce
the MSE (Mean Square Error) of its results relatively to the
expected results. Therefore, as new parameters are included in the
BP formulas, GA Trainer 160 creates new mutations. These formula
mutations guarantee a continuous, real time mechanism, which
dynamically responds and improves the GA's calculations.
[0051] Using these relationships parameters and their widthwise
relative grades, GA Profiler 150 maps the entire users' objects'
psycho social personal patterns. Then, it can indicate the relative
patterns in each sub-segment. The profile mapping method is
actually a definition of the way that the different Nr parameters
and their derived instances are joined together. As mentioned,
there are many types of those variables, and their specific
combination may be complex. For example: contradictive trends of
Relation Level and Aggression Level of the analyzed user's entity,
over time, may indicate a submission of one of the parties. The
correlation between these mapping results and the actual reality
states are determined by GA-Trainer 160. Then, the final step of
the profiling process is one or more combinations (or relationship
parameters mapping) that defines the specific behavioral parameter
level.
[0052] GA Profiler 150 performs and analysis of Behavioral Pattern
(BP) variables 180 which are the formal representation of user
archetypes 80 (see FIG. 6 below). BP variables 180 represent
specific human characteristics or sets of characteristics. GA
Profiler 150 receives external profile data from profiling studies
165 via external BP mapping module 169. GA trainer 160 receives
external profile data from profiling studies 165 via a profile
analysis manager 166. Profile analysis manager 166 may comprise an
automated self-learning module which is responsible of the BP's
formulas composition. Based on past formulas and data, it chooses
the best NR's and normalization environments in order to create
formula which best fits each BP.
[0053] Results of GA Profiler 150 may be user to generate queries
by a query calculator 230, for example query calculator 230 may be
arranged to generate different behavior patterns categories.
Therefore, it can point to which categories a BP relates.
Consequently, category-BP segmentation level is created and is
being used in broad category persona definition. Query calculator
230 may thus be used to segment users of an application such as a
social network, for different purposes.
[0054] In certain embodiments, BP Variables 180 are composed by the
following stages. GA Profiler 150 generates behavior pattern
categories using the base variables, called BPs. Every BP is
composed from a closed, pre-defined group of Nr variables (every Nr
variable can be used for a couple of different BPs)--managed by an
External BP Mapping Module 169 that receives profiling data from
profiling studies 165.
[0055] There are several object levels that contain BP variables. A
single BP can be produced by Nr variables from the same
data-object, in lower levels, other BP variables, from the same
object level but from lower a data-hierarchical order or any
combination of them. The composition order of a BP variable is a
dynamic value related combination.
[0056] The exact combination for each and every BP variable is
determined using the ANN Trainer, and stored in the BP Composition
Table--ANN Matrix1 155. The behavior pattern variable is defined by
the following syntax: BP(S) i: [0057] The `S` stands for the
definition of the source data entity. [0058] The `cl` stands for
the definition of the current aggregation level. [0059] The `i`
stands for the definition of the specific variable in that
series.
[0060] The transition for each profile level is combined from two
vectors--the direct lower profile data and its parallel
(normalized) gross data level.
[0061] FIG. 6 is a high level schematic illustration of the
behavior patterns building process, according to some embodiments
of the invention.
[0062] In certain embodiments, analysis unit 140 generates an
internal mapping 142 of BP variables 180 from ANN Matrix 155 and
presents internal mapping 142 and BP external mapping 169 from BP
modeling manager 166 (see FIG. 4) to a pattern analyzer 146.
Pattern analyzer 146 is managed by analyzer manager 144 (may be
part of profile analysis manager 166) that receives gross data 115
and controls a semantic analyzer 148 that uses a semantic
dictionary 143. The pattern analysis and the semantic analysis, as
well as BP transformation matrix 172 (exemplified above) are
combined and inputted into a Profile Dynamic Level Transformation
(PDLT) calculator 174 which derives BP variables 180.
[0063] In certain embodiments, BP variables derivation is carried
out as follows. Profile variables (BPs) 180 are generated in
dynamic transaction mechanisms.
[0064] The correlation between each and every variable between
levels is determined by PDLT calculator 174. This mechanism defines
a list of transformation types. Unlike the GDLT (for the gross
variables), the PDLTs holds a definition of combination functions
fBP that define the way to create each of the BP variables. The fBP
is using a set of pre-defined transformation relations T.sub.t<
> (for the aggregative BPs) and a coefficient parameter for each
organ.
[0065] In certain embodiments, profiling and segmentation data 175
of the users may be used for different aims such as advertising and
e-commerce. For example, an advertisement managing unit may be
arranged to generate advertisements relating to user archetypes 80
and a campaign managing unit may be arranged to present the
generated advertisements to users of e.g. a social network platform
according to their association with user archetypes 80. In another
example, a proposal generator may be arranged to generate
commercial proposals relating to user archetypes 80 and a commerce
manager may be arranged to present the generated commercial
proposals to users of e.g. a social network platform according to
their association with user archetypes 80. The commerce manager may
be further arranged to manage electronic commerce of the users in
relation to their associated user archetypes 80.
[0066] FIG. 7 is a high level schematic flowchart of a profiling
method 300 according to some embodiments of the invention. At least
one stage of method 300 is at least partially carried out by at
least one computer processor.
[0067] Method 300 comprises the following stages: Receiving user
activity data (stage 310); deriving statistical data that
characterizes the user activity data (stage 320); normalizing the
statistical data related to each user with respect to user
population(s) (stage 330) and analyzing a correspondence between
normalized user study data and user archetypes (stage 340).
[0068] Method 300 may further comprise representing the relation
between the normalized user study data and the user archetypes by
an artificial neural network (stage 350).
[0069] Analyzing the correspondence (stage 340) may comprise
applying a heuristic genetic algorithm on the artificial neural
network (stage 360); training the heuristic genetic algorithm with
obtained normalized user study data (stage 362); and associating,
for each user, the normalized statistical data with one of the user
archetypes (stage 370).
[0070] Method 300 may further comprise presenting the association
of users and user archetypes to an application (stage 380) and
profiling users of a social network (stage 382).
[0071] Certain embodiments of the invention comprise a computer
program product comprising a computer readable storage medium
having computer readable program embodied therewith. The computer
readable program may comprise computer readable program configured
to implements any of the stages in method 300.
[0072] In certain embodiments, profiling system 100 may be used as
an advertising management, targeting and media buying platform.
Based on the behavioral psycho-social engine, profiling system 100
is arranged to provide a unique and simple way to plan, provision,
and test and easily manage social networks Ads campaigns.
[0073] Advantageously, profiling system 100 and method 300 may be
designed for the direct users' engagement layer, mainly on social
platforms, using display ads from social inventory. Profiling
system 100 and method 300 take different types of raw data from
social networks and based on the unique analytical processes,
accurately differentiates the users by analyzing their
psycho-social behavioral pattern. Therefore, marketing messages can
be more finely-tuned and personally directed to each psychological
persona type.
[0074] In addition to the current social network methods of
providing advertisers with obvious data such as user's interests,
groups, geographical location and etc., profiling system 100 and
method 300 further delve into yet another layer of user information
that determines users' personalities. Profiling system 100 and
method 300 accurately map valuable data from multi-layered virtual
communication in social networks and create users' personal,
archetypes-based customer profile. Advantageously, profiling system
100 and method 300 target the exact human profile user group for
personalized user engagement and may split campaigns using an
accurate users' clustering--reaching every different customer type
with a designated, relevant advertising massage.
[0075] In the above description, an embodiment is an example or
implementation of the invention. The various appearances of "one
embodiment", "an embodiment", "certain embodiments" or "some
embodiments" do not necessarily all refer to the same
embodiments.
[0076] Although various features of the invention may be described
in the context of a single embodiment, the features may also be
provided separately or in any suitable combination. Conversely,
although the invention may be described herein in the context of
separate embodiments for clarity, the invention may also be
implemented in a single embodiment.
[0077] Certain embodiments of the invention may include features
from different embodiments disclosed above, and certain embodiments
may incorporate elements from other embodiments disclosed above.
The disclosure of elements of the invention in the context of a
specific embodiment is not to be taken as limiting their used in
the specific embodiment alone.
[0078] Furthermore, it is to be understood that the invention can
be carried out or practiced in various ways and that the invention
can be implemented in embodiments other than the ones outlined in
the description above.
[0079] The invention is not limited to those diagrams or to the
corresponding descriptions. For example, flow need not move through
each illustrated box or state, or in exactly the same order as
illustrated and described.
[0080] Meanings of technical and scientific terms used herein are
to be commonly understood as by one of ordinary skill in the an to
which the invention belongs, unless otherwise defined.
[0081] While the invention has been described with respect to a
limited number of embodiments, these should not be construed as
limitations on the scope of the invention, but rather as
exemplifications of some of the preferred embodiments. Other
possible variations, modifications, and applications are also
within the scope of the invention. Accordingly, the scope of the
invention should not be limited by what has thus far been
described, but by the appended claims and their legal
equivalents.
* * * * *